I have no idea why this might be limited by the light source being collimated?
I mean, you can get electricity from PV illuminated by a laser, and everything I've heard so far says it's easier than with sunlight because you can match the frequency of the laser to the band gap of the PV.
It sounds like you're mistaking PV for a thermal system.
In a PV cell, you have a semiconductor. Semiconductors have this thing called a "band gap", which is the energy needed to get an electron from the valence band to the conduction band: https://en.wikipedia.org/wiki/Band_gap
The limits to efficiency of a solar panel is that sunlight has photons of many energy levels; the photons with energy less than the band gap do nothing, those with more, waste the excess.
A laser can have energy tuned to this band gap, at which point the PV part becomes ~99.9% efficient. (The laser part is not close to that efficiency).
I'm not talking about a thermal system, I'm talking about having to deal with the thermals of your inefficiencies. That energy that doesn't get converted to electricity is converted to heat. And you have to deal with it.
The type of laser based PV that you're taking about that's highly tuned is at maximum 27% efficient. Not 99%.
That's a 73% waste you have to manage
I don't know where you get these ideas from, but they're very very wrong.
You can already buy PV for sunlight with higher efficiency than 27%. Heck, even with a single band gap the limit for *sunlight* with all the problems I've just stated (because a broad spectrum will waste energy and the sun's spectrum is very broad) still gets you to the 33% Shockley–Queisser limit for one junction on the solar spectrum: https://en.wikipedia.org/wiki/Shockley–Queisser_limit
But for monochromatic light, tuned to the band gap, stuff people have already built is several times better than that, and the theoretical limit is basically how finely you can tune the laser bandwidth and how precisely you can control impurities that broaden the band gap.
Here's one designed for monochromatic laser use, that was actually built, with results published in early 2019, with 70% where they recon doing it better would get to 80%: https://krichlab.ca/wp-content/uploads/2019/02/Xia18-PVSC.pd...
As with all quantum systems, which is what both PV and lasers are, the ultimate efficiency of a tuned laser-and-band-gap pair is as good as your engineering, hence ~99.9% (on the cell side) if you control absolutely all parts of it properly.
Sure, you absolutely can do it. But material science quickly becomes a major limit.
For something 15% efficient like a high quality PV cell, for every 100 watts you want to be usable on the receiving side, the receiver has to bleed off 566 watts of heat. And that's 566 watts of waste heat that is highly concentrated.
Consider a single residental power circuit. 12A maximum, 120v, that's 1440 watts at delivery. For PV power delivery via laser, that PV would need to dissipate 8 kilowatts of waste heat. One a very small surface